Understanding Formative vs Reflective Indicators in SEM

[@iqbalkarim7107]

Thanks for explaining. It get me easy to go.

[@bigcat1993]

Very helpful video

[@LifeIsBayWatch]

Thanks for sharing, very helpful!

[@alexbarus1452]

thanks help me much to understand formative second order cons...

[@chefberrypassionateresearcher]

Sir, please tell how to do Invariance of Formative Measures (Structure invariance, Slope invariance, and Residual invariance)? Please share a link of the video if u have one...

[@shivigupta29]

Thank you for the insightful presentation. In the case of family social capital and international entrepreneurial orientation, is it reflective or formative?

[@joelcollier9387]

My own personal research has not had to address those constructs...so I am not familiar with those literature streams. If there are multiple dimensions that make up family social capital, then it is a formative construct. Same with Intern. Entrepr. Orientation. Hope that helps.

[@thinhnguyen5180]

Thank you so much, your video is really helpful for me. Could you make a video on processing 2nd order constructs using AMOS, please?

[@joelcollier9387]

Thanks for the kind words. I have a pretty large section on how to run 2nd order constructs in my book. I have a couple other topics that are lined up first but that one will definitely be addressed in the future.

[@thinhnguyen5180]

@@joelcollier9387 Thank you so much, I look forward to watching your videos.

[@yuqiliu3285]

Dr. Joel, thanks for the video! Do you think PLS SEM can handle 'measurement errors'? What about SPSS PROCESS? Is PROCESS a good option when we have formative constructs? Thank you!

[@joelcollier9387]

Can PLS handle measurement errors? Yes, but I don't recommend PLS there is so many issues with that program. In my book, I actively tell people to avoid that program. It is not a covariance based program. PROCESS is great but it will not handle formative constructs. It will also only handle one DV at a time which can be problematic. AMOS is my preferred choice for formative indicators but MPLUS does a nice job with it as well.

[@ilarefiana5131]

Thank you for the presentation. In the case of servqual measurement, is it a reflective or formative? Thank you in advance 🙏

[@joelcollier9387]

It is definitely formative. I wrote a paper 2006 in the Journal of Service Research that specifically talks about why E-service Quality is a formative higher order construct

[@ilarefiana5131]

@@joelcollier9387 I have just read your paper. We'll cite it in our research. Thank you 🙏

[@inspiredliving4279]

Sir, please tell me, if we use a psychometric assessment ( say scale of self compassion) which has six dimensions to it and we wish to see its effect on an endogenous variable which also is a psychological scale say psychological well-being which is measured by six dimensions of its own . Then , would it not be obvious to use formative model as the scales are already established and dimensions have been confirmed to be capturing the unobservable construct fully?

[@joelcollier9387]

So your question is a little more complicated than just a yes/no answer. If your independent variable is a higher order construct that has 6 dimensions, then yes, it needs to be treated as a formative construct. The complication comes in testing a relationship from a higher order construct to another higher order construct. With a higher order dependent variable, the relationship (path) from the independent variable must go to the first order dimensions...not the higher order construct. The reason is the first order dimensions make up the higher order construct and thus, it must influence the first order constructs. In my book, I go into great detail about this and give an example. It would take way too much space to discuss it in the comments. I encourage you to check it out if you are looking for more information on higher order constructs and their relationships to variables.